DATAS / Documentation
Documentation
Put DATAS to work
How to prepare the two DATAS files, check them, and run the free dashboard, plus the tax credit crosswalk and policy language. For governance, the roadmap, and common questions, see About DATAS.
On this page
Overview
DATAS (Data Alignment and Tutoring Assessment Standards) is an open, MIT-licensed specification for describing high-dosage tutoring programs in a machine-readable, privacy-preserving way.
It makes tutoring data interoperable across schools, districts, states, providers, platforms, scholarship organizations, and researchers, without requiring any party to transmit student-level records.
The standard is at v1.0.0. Twenty-nine agencies covering more than 931,000 students have committed to it, Louisiana has adopted model regulations that reference it, and a twelve-person steering committee stewards it.
Everything here, including the dashboard, the schema, and this website, is open source. Inspect, fork, or extend it at github.com/accelerate-usa/tutor-data-standard.
In short. Export your tutoring data into two files that match the schema, validate them, and share aggregate outputs. Why it matters: the problem DATAS solves.
Design principles
- Lowest common denominator. The schema is intentionally small. If it exists in most provider systems today, we want it. If it doesn't, we don't ask for it.
- Plain files first. CSV, Excel, and JSON, the formats that survive the trip through finance and IT. No SDKs required.
- On-device by default. Student-level records never have to leave the environment that produces them. Outputs are aggregate-only.
- Open and stewarded. MIT license. Any contributor with an accepted pull request is invited to request a committee seat.
Getting started
Most teams get up and running by committing to interoperability, preparing the two DATAS files, and validating a sample early. Adoption typically takes a district about ten hours spread over two weeks, with under an hour a week of upkeep after that.
If you're a district
Ask providers for DATAS-formatted tutoring activity, keep the district-held student file on your side, run the dashboard with both files, and use the results for oversight and implementation improvement.
If you're a provider
Export DATAS-formatted tutoring activity, send that package to the district, and work with the district on implementation insight without taking possession of district student data.
-
Commit to interoperability
Make one shared commitment before anyone exports, requests, or validates data. Put DATAS in the working agreement, provider conversation, data request, or contract language so everyone is aiming at one format from the start. When you're ready to signal your support, add your organization to the commitment.
-
Prepare the DATAS files
Keep the standard path simple: the provider prepares
sessions.csv(who was served, when, for how long, by whom, and in which subject) and sends it to the district. The district keepsstudents.csv(identifiers, demographics, program flags, and assessment history). Providers don't receive district student data in the standard path.The dashboard combines the two files after upload, so the main job is matching the format and keeping
student_idvalues consistent. Compare your files to the sample files and the schema early. -
Validate with the tool
Open the hosted dashboard, or run the same app locally if records can't leave your environment. Upload both files; the app starts once both are present. It checks required columns, data types, missing values, row-level quality, and whether
student_idlines up across the files.The dashboard still loads data with warnings, so treat warnings as cleanup work. A chart that loads isn't the same as a clean file. Providers should clean
sessions.csvagainst the same definitions before delivery, then review the district's two-file pass together. -
Improve the results
DATAS pays off when it becomes routine: one standard export, one district-held student file, one repeatable validation habit. Build it into RFP or CFP language, contracts, and guidance, and use each cycle to look at dosage, access, equity, outcomes, and cost, then make one real change.
The files
Schema
The data dictionary on GitHub is the authoritative field list, generated from the schema registry. The tables below are a quick reference. Download the sample files to see them populated.
sessions.csv
Prepared by the provider. One row per student per session.
| Field | Type | Required | Description |
|---|---|---|---|
student_id | identifier | Yes | Matches students.csv exactly after trimming whitespace. Stored as text. |
session_topic | categorical | Yes | Subject. Common aliases such as "reading" and "mathematics" are normalized; current analysis expects math or ela. |
session_date | datetime | Yes | Date or date and time of the session, in any parseable format. |
session_duration | numeric | Yes | Length of the session in minutes. Non-negative. |
session_ratio | string | Recommended | Student-to-tutor ratio, such as 1:1 or 1:3. |
tutor_id | identifier | Yes | Stable identifier for the tutor. |
students.csv
Kept by the district. One row per student.
| Field | Type | Required | Description |
|---|---|---|---|
student_id | identifier | Yes | Join key shared with sessions.csv. Stored as text, so leading zeros survive. |
district_id / district_name | identifier / string | Yes | District identifier (stored as text) and name. |
school_id / school_name | identifier / string | Yes | School identifier (stored as text) and name. |
current_grade_level | integer | Yes | Grade served. Accepts K and Pre-K labels; −1 through 12 after normalization. |
gender | categorical | Recommended | As reported locally, for example Female, Male, Nonbinary, Unknown. |
ethnicity | categorical | Recommended | As reported locally. Keep categories compact; the dashboard warns past ten. |
ell / iep / economic_disadvantage | Yes / No / Unknown | Recommended | Blank values stay unknown; they aren't converted to No. |
gifted_flag / homeless_flag / disability | Yes / No / Unknown | Recommended | Optional student-context flags. |
ela_outcome_measure_1–3 | numeric | Recommended | Ordered ELA outcome measures, earliest to most recent. |
math_outcome_measure_1–3 | numeric | Recommended | Ordered math outcome measures, earliest to most recent. |
performance_level_earliest / previous / most_recent | categorical | Recommended | Optional proficiency labels aligned to the outcome sequence. |
Outcome note. DATAS doesn't require state tests. Any ordered numeric outcome sequence works, as long as the measures are in order and comparable within subject. Gain and value-added views assume the three ELA measures share one scale and the three math measures share another.
Upload note. The standard exchange format is sessions.csv plus students.csv, but the app also ingests .xlsx, .xls, and .json files and normalizes supported aliases.
Validation rules
The validator checks whether your files can support the analysis modules, keeps identifiers as text, applies supported column aliases, and shows normalization details before you trust the charts.
| Check | What it looks for |
|---|---|
| Structure | Required columns are present. Canonical names are preferred; a limited set of documented aliases is accepted and reported. |
| Types | IDs, dates, numbers, grades, and subgroup flags are parsed explicitly instead of left to spreadsheet guessing. |
| Coverage | Missing fields and values are surfaced. Partial datasets often still load, with warnings that show what's usable. |
| Joins | Every tutoring record points to a student record. Unmatched sessions can't be analyzed, so join problems are reported directly. |
| Range | Dates parse, durations are positive, and subject labels are recognizable, so subject-level analyses can run. |
The upload screen also shows what each upload supports, a normalization summary, and optional local mapping profiles for recurring non-canonical uploads.
The dashboard
The toolkit
The DATAS dashboard turns your tutoring data into live visibility on dosage, coverage, and fidelity. It's free and open source. Use the hosted version for a quick check, or run it on your own machine so nothing leaves your environment.
Upload your DATAS files
Use
sessions.csvandstudents.csv. The app also accepts.xlsx,.xls, and.json, and normalizes supported column aliases.Read the report
Five checks in plain English. Blocking errors are flagged in red and minor issues in amber.
Export the aggregate
A shareable summary with no student-level records, for assessing implementation and improving outcomes.
What's bundled
- DATAS dashboard: the validator and live insight view.
- Data dictionary: the canonical spec.
- Sample files: realistic, synthetic students and sessions.
- Model state policy: Accelerate's 2025 policy framework for state tutoring programs, which points to DATAS.
- Source on GitHub: fork it, extend it, ship new charts.
See what each view shows on the homepage.
Running the tools locally
The dashboard is a Python Streamlit app. To run it on your own machine:
- Install Python 3.11 or newer.
- Clone the repo from GitHub.
- Install the toolkit's dependencies and launch the app.
Terminal (bash or zsh)
git clone https://github.com/accelerate-usa/tutor-data-standard.git
cd tutor-data-standard
python3 -m venv .venv
source .venv/bin/activate
pip install -r toolkit/requirements.txt
streamlit run toolkit/descriptives.py
# the dashboard opens at http://localhost:8501
PowerShell
git clone https://github.com/accelerate-usa/tutor-data-standard.git
cd tutor-data-standard
py -m venv .venv
.venv\Scripts\Activate.ps1
pip install -r toolkit/requirements.txt
streamlit run toolkit/descriptives.py
# the dashboard opens at http://localhost:8501
Because the code is open source, you can extend the app. If you build something broadly useful, open a pull request.
When you report publicly, cite the version you validated against ([email protected]).
Policy and funding
The Federal Scholarship Tax Credit
Starting January 1, 2027, the Federal Scholarship Tax Credit gives taxpayers a federal credit for donating to Scholarship Granting Organizations (SGOs), which fund K–12 scholarships. Academic tutoring is named in the law as an eligible expense, and Treasury estimates about $26 billion a year will flow through the program by 2030.
Under Treasury's proposed rules, tutoring for students who qualify through the rules' safe harbor needs an independent audit confirming that tutors were qualified, that impact was assessed, and that services met "a duration and frequency commonly required to produce benefit." That audit runs on session-level records, and DATAS gives every SGO, school, and provider the same free format for them.
The money is tracked. Now the tutoring can be too.
Crosswalk
| The audit needs | DATAS fields and tools | Status |
|---|---|---|
| Duration and frequency | session_date and session_duration on every session; the dashboard reports minutes per student per week. | Covered |
| Proof of delivery | Each session is a dated, timed record tied to student_id and tutor_id, so sessions billed can be checked against sessions delivered. | Covered |
| Assessed impact | ela_outcome_measure_1–3 and math_outcome_measure_1–3, on any comparable scale, with gain and value-added views. | Covered |
| Group size | session_ratio records how many students each tutor worked with. | Covered |
| Student need | Baseline outcome measures plus ell, iep, and economic_disadvantage show who was served. How a student's need was identified isn't recorded. | Partly |
| Qualified tutors | tutor_id ties every session to a tutor. Tutor qualifications aren't part of the standard yet. | Not yet |
| Student privacy | The tools run inside your own environment, so student records never have to move to prove anything. | Covered |
Who it helps
SGOs
Confirm sessions were delivered before you pay, and keep records that are ready for an audit.
Auditors
Read one standard evidence format instead of a different layout from every SGO.
Payment platforms
Give tutoring vendors a ready-made data spec to meet before they're paid.
How an SGO can use DATAS
- Require DATAS-formatted
sessions.csvfiles from the tutoring providers you pay. - Validate each file, and compare the sessions delivered to the sessions invoiced before you pay.
- Keep the validated files and the aggregate report as the record for your annual audit.
A note on the rules. This crosswalk is based on Treasury's proposed rules released October 1, 2026, which may change before they're final. Treasury also plans separate guidance on which tutoring expenses qualify. Read Accelerate's statement on the proposed rules.
Policy language
State and district leaders can require DATAS with a single clause.
Model RFP clause. "Vendor shall produce quarterly tutoring program data in conformance with the DATAS standard (datas@>=1.0.0) as validated by the DATAS dashboard. Aggregate reports shall be made available to the district within 14 days of quarter close."
Louisiana adopted a version of this language in 2025.
Model state policy
For a full statewide framework, Accelerate's 2025 Model State Policy for High-Dosage Tutoring covers definitions, LEA and SEA duties, a state-vetted list of tutoring providers, and annual reporting. Its reporting section (Section 7) points states to DATAS for best practices in data collection and use.
